“AI-generated misinformation became increasingly widespread, as I warned in December 2022, but the effects have – thus far — been less than I feared. I continue to be quite concerned, and think vigilance continues to be required.”
Our reading, the sentence that was checked:
AI-generated misinformation became increasingly widespread, but its effects have thus far been less than Marcus feared; he remains quite concerned and thinks vigilance continues to be required.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, 25 AI Predictions for 2025, from Marcus on AI, paragraph 13.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-25-ai-predictions-for-2025-from-marcus-570d43-a2
We recorded 37 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2025-01-01. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- thus far (qualifier, stated by the speaker)
- less than I feared (qualifier, stated by the speaker)
From the speaker
- AI-generated misinformation became increasingly widespread
- as I warned in December 2022
- the effects have – thus far — been less than I feared
- vigilance continues to be required
Our interpretation
- The retrospective comparison runs from after the December 2022 warning through the end of 2024.
- Increasingly widespread means a broader observed distribution of AI-generated false information, not necessarily increased persuasive effectiveness.
- Counts of unreliable AI-generated sites are an imperfect indicator: not every article on such sites is false, and discovery counts can rise with monitoring effort.
- The comparison with Marcus's fears requires an identifiable prior impact expectation. Concern and the recommendation for vigilance are personal or normative statements.
Counts as supported if: Comparable monitoring establishes wider distribution through 2024, and a documented prior expectation permits evidence of actual effects to establish that those effects were smaller.
Counts as contradicted if: Comparable monitoring establishes no growth or a decline in distribution, or measured effects meet or exceed a documented prior expectation.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
NewsGuard reported 49 identified unreliable AI-generated news sites in May 2023 and more than 600 in December 2023 [e6]. The Decoder reported NewsGuard's count of 966 sites on July 6, 2024 [e7]. This supports growth in identified sites through July 2024, but does not establish a comparable prevalence trend through year-end [e6, e7]. TIME's October 30, 2024 report described political deepfakes as a small part of wider misinformation campaigns while noting uncertainty about their effects [e8]. These records do not establish effects through December 2024 or a documented baseline for Marcus's fears, so the full comparison remains unresolved [e6, e7, e8].
Limitations
- Site discovery counts do not establish audience exposure, the proportion of content that is false, or how much growth reflects monitoring coverage [e6, e7].
- TIME's reporting preceded the U.S. election and does not measure all worldwide harms [e8].
- The packet provides no measurable prior expectation against which to compare actual effects [e6, e7, e8].
Evidence (3 items)
Supports the claim: NewsGuard,
more than 600
Why it matters here: The report describes growth from 49 identified domains in May to more than 600 in December and documents AI-assisted false narratives. This supports expanding observed distribution, subject to the limits of site discovery as a prevalence measure.
- Publisher
- NewsGuard
- Kind of source
- Independent measurement
- Describes events of
- Public from
Supports the claim: The Decoder,
966
Why it matters here: The later count is consistent with continued expansion of the identified unreliable-site population, and the article describes recurring false claims. This is secondary reporting of NewsGuard's findings, not an independent replication or a measure of audience persuasion.
- Publisher
- The Decoder
- Kind of source
- News report
- Describes events of
- Public from
Context only: TIME,
a small part
Why it matters here: The report provides pre-election context consistent with limited observed impact, while noting serious measurement gaps. It cannot establish final election effects, worldwide effects or their size relative to Marcus's private fears.
- Publisher
- TIME
- Kind of source
- News report
- Describes events of
- Public from
What would settle it
A documented prior expectation specifying harms and their anticipated scale, measurements of those harms through December 2024, and comparable prevalence evidence through year-end that accounts for changes in monitoring coverage.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, 25 AI Predictions for 2025, from Marcus on AI
AI-generated misinformation became increasingly widespread, as I warned in December 2022, but the effects have – thus far — been less than I feared. I continue to be quite concerned, and think vigilance continues to be required.
- Where
- paragraph 13
- Published
- 2025-01-01
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-25-ai-predictions-for-2025-from-marcus-570d43-a2: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-25-ai-predictions-for-2025-from-marcus-570d43-a2.